In particle physics, the likelihood ratio ordering principle is frequently used to determine confidence regions. This method has statistical properties that are superior to that of other confidence regions. But it often requires intensive computations involving thousands of toy Monte Carlo datasets. The original paper by Feldman and Cousins contains a recipe to perform the toy MC computation. In this note, we explain their recipe in a more algorithmic way, show its connection to 1-CL plots, and apply it to simple Gaussian situations with boundaries.